Targeted chemotherapy by intratumour injection of ... - Nature

Gene Therapy (1998) 5, 1070–1078  1998 Stockton Press All rights reserved 0969-7128/98 $12.00 http://www.stockton-press.co.uk/gt

Targeted chemotherapy by intratumour injection of encapsulated cells engineered to produce CYP2B1, an ifosfamide activating cytochrome P450 M Lo¨hr1, P Mu¨ller1,2 P Karle3,4, J Stange1, S Mitzner1, R Jesnowski1, H Nizze5, B Nebe1, S Liebe1, B Salmons2 and WH Gu¨nzburg3,4 Departments of 1Medicine and 5Pathology, University of Rostock, Germany; 2Bavarian Nordic Research Institute, Martinsried, Munich, Germany; 3GSF-Neuherberg, Oberschleissheim, Germany; and 4Institute of Virology, University of Veterinary Sciences, Vienna, Austria

The prognosis of pancreatic adenocarcinoma is poor and current treatment ineffective. A novel treatment strategy is described here using a mouse model system for pancreatic cancer. Cells that have been genetically modified to express the cytochrome P450 2B1 enzyme are encapsulated in cellulose sulphate and implanted into pre-established tumours derived from human pancreatic cells. Cytochrome P450 2B1 converts the chemotherapeutic

agent ifosfamide to toxic metabolites. Administration of ifosfamide to tumour-bearing mice that were recipients of implanted encapsulated cells results in partial or even complete tumour ablation. These results suggest that in situ chemotherapy with genetically modified cells in an immunoprotected environment may prove useful for application in man.

Keywords: cytochrome P450; ifosfamide; encapsulation; prodrug therapy; pancreatic cancer; local therapy

Introduction The effectivity of chemotherapeutic intervention for solid tumours is limited by: (1) access to the tumour; (2) the degree of drug sensitivity of the tumour; and (3) local and systemic toxicity of the chemotherapeutic agent.1 Furthermore, resistance to treatment, both on a molecular (MDR gene)2 and multicellular3 level, are often encountered. In addition to these general problems, human pancreatic adenocarcinoma is particularly aggressive, has a poor prognosis and at the time of diagnosis, the tumour is generally in an advanced stage no longer suitable for resection.4 Even if resection can be undertaken, recurrence at the original site of the tumour shortens survival.5,6 Despite recent developments,7,8 conventional chemotherapy, using drugs such as cisplatin, 5-fluorouracil, taxol and more recently gemcitabine, has had limited success in the clinic since the necessary high local drug doses cannot be achieved without significant systemic toxicity.3,5,8 Novel chemotherapeutic prodrugs might offer a way round these toxicity problems if they could be activated locally. Among such prodrugs, ifosfamide,9,10 has given promising results in vitro.11 Ifosfamide is normally converted in the liver to tumoricidal metabolites by the liver active enzyme cytochrome P450 2B1 (CYP2B1).12 The metabolites phosphoramide mustard, which alkylates DNA, and acrolein, which alkylates pro-

teins, are then systemically distributed via the blood to finally reach the tumour.13,14 Although DNA alkylation and crosslinking occurs in many cells, its cytotoxic effects are limited to dividing cells such as tumour cells, but also other replicating cells such as haematopoietic cells. It is this systemic toxicity that currently limits the dose of ifosfamide that can be applied.3 If these problems could be overcome ifosfamide would be a particularly interesting prodrug for a variety of oncological applications since, in contrast to the prodrug acyclovir (or ganciclovir) which is activated by the herpes simplex virus thymidine kinase gene product, ifosfamide is an approved anticancer agent, supported by many years of experience in the clinic. Further, the cytotoxic effects of ifosfamide are not, as is the case in the HSVtk/gancylovir system, dependent on the presence of a tight cell–cell contact over gap junctions for a bystander effect.15 In this report, an established animal model for pancreatic adenocarcinoma16,17 has been used for local (rather than systemic) chemotherapy with ifosfamide by delivering the activating enzyme CYP2B1 in an immunoprotected environment directly into the tumour. Local conversion of ifosfamide in the tumour produces high active metabolite concentrations and limits systemic toxicity.

Results Correspondence: WH Gu¨nzburg, Institute of Virology, University of Veterinary Sciences, Josef-Baumanngasse 1, A-1210 Wien, Austria Received 2 October 1997; accepted 10 February 1998

Characterisation of cytochrome P450 2B1-expressing cells Feline kidney cells18 carrying an expression vector, in which the CYP2B1 cDNA is placed under the transcrip-

Targeted chemotherapy using encapsulated cells M Lo¨hr et al

tional control of the CMV immediate–early promoter (Figure 1a), were analysed for expression of functional CYP2B1. This enzyme specifically dealkylates 7-pentoxyresorufin to the fluorescent compound, resorufin.19 Stably transfected CYP2B1 cells were shown to display CYP2B1 enzymatic activity proportional to the cell number analysed. Typically one of the cell clones produces around 2 pmol/105 cells. Cells expressing CYP2B1 convert ifosfa-

mide to metabolites (phosphoramide mustard and acrolein) which are cytotoxic for dividing cells.13,14 This drug-induced suicide effect could be first observed 2 days after the administration of pharmacologically active concentrations of ifosfamide and were clearly visible after 6 days (Figure 1b; 0.25–2 mm). A value of 0.25 mm is in the range of plasma drug concentrations (0.1–0.5 mm) observed after administration of the drug to patients.20 In contrast, cells not expressing CYP2B1 were affected to a much lesser extent (72.7% ± 9.8 viability) than CYP2B1expressing cells (9.6% ± 1.8 viability) at pharmacological concentrations (0.25 mm) of ifosfamide (Figure 1b). To verify that this concentration is also pharmacologically relevant in mice, seven mice were treated with 100 mg ifosfamide/kg and the plasma levels assayed. The mean plasma concentration after 15 min was 0.65 mm (range 0.55–0.88) and after 90 min 0.34 mm (range 0.21–0.53), suggesting a half-life of 78.8 min.

Bystander cytotoxic effect Co-cultivation of CYP2B1-expressing and nonexpressing cells demonstrated that nonexpressing cells could also be killed by the toxic metabolites of ifosfamide produced by the CYP2B1-expressing cells (data not shown). The ability of genetically modified cells that activate a prodrug also to affect surrounding nonmodified cells has been termed a bystander effect.21 To investigate whether this bystander effect is dependent on direct cell-to-cell contact, as is the case for the HSVtk-mediated effect15 or whether it is mediated by a freely diffusable metabolite, cells expressing CYP2B1 and nonexpressing reporter cells were separated by a filter with 0.45 ␮m pores. To simulate the situation during treatment of a pancreatic tumour, a human pancreatic tumour-derived cell line, PANC117,22 was used. The combination of non-CYP2B1expressing cells and PANC1 cells showed no significant change in cell number or viability when treated with pharmacologically active ifosfamide concentrations (Figure 1c). At 5 mm, CYP2B1 independent nonspecific cytotoxicity of ifosfamide was observed. In contrast, the PANC1 cells that were seeded together with CYP2B1expressing cells showed a significant decrease in the survival of PANC1 (and CYP2B1-expressing) cells at the pharmacological dose of 0.25 mm ifosfamide. This bystander effect became more pronounced with increasing concentrations of ifosfamide. To rule out that this effect is specific to the PANC 1 cells, a number of other cell lines (epithelial cells of human and feline origin, insulinoma and B cells) were also tested and similar results were obtained (data not shown). Figure 1 Construction and biological activity of a cytochrome P450 2B1 expression plasmid. (a) Structure of the CYP2B1 expression construct: the cytochrome P450 2B1 is expressed from the cytomegalovirus (CMV) promoter and polyadenylated at an SV40 polyA signal (pA+). The plasmid also carries an SV40-neomycin resistance gene (neo) −pA+ cassette allowing selection in G418-containing media. (b) Differential killing of CYP2B1 expressing cells (쐽) compared with non-expressing cells (왖) after treatment with various concentrations of ifosfamide. The percentage of cells surviving after 1 week of treatment are shown. Error bars denote standard error of the mean of triplicate determinations. (c) Dose-dependent cytotoxic bystander effect of CYP2B1 expressing cells (왖) compared with non-expressing cells (쐽) on PANC1 cells. The PANC1 target cells were separated from the CYP2B1 expressing or non-expressing cells by a 0.45 ␮m filter. The percentage of PANC1 cells surviving after 1 week of treatment are shown.

Encapsulation of CYP2B1-expressing cells and characterisation The aim of this study was to inject CYP2B1 expressing cells into tumours of mice before treatment with ifosfamide to achieve local killing of the tumour. The CYP2B1 expressing cells were encapsulated in cellulose sulphate23 for immobilisation in the tumour and for protection from the immune system. Cellulose-sulphate capsules of 200– 500 ␮m diameter were produced which pass comfortably through a 21 G needle (inner diameter 0.6 mm). The integrity of empty and cell filled capsules after flushing through a 19 G needle with a 1 ml standard syringe was confirmed by phase contrast and electron microscopy. After repeated passing through the needle, the capsules

1071


1072

were also placed in tissue culture for several weeks. No outgrowth of cells from these capsules was observed during the 6-week observation period, indicating that the capsules were still intact (data not shown). The vitality of the encapsulated cells was assessed by Life&Dead assay (MobiTec, Braunschweig, Germany) and confocal laser microscopy. Cells at the outer edge of the capsules tended to survive longer than those in the center of the capsules (Figure 2a). Approximately 70% of the cells were alive 2 weeks after encapsulation, and

around 50% 4 weeks after encapsulation. Typically, a 200 ␮m capsule contains up to 104 cells and single capsules could be shown to yield 0.15 pmol CYP2B1 by the resorufin assay. This compares favourably with the levels detected from these cells in normal culture conditions (ie approximately 2 pmol/105 cells). Further, enzyme activity could be detected up to 4 weeks after encapsulation of transfected cells, confirming that the cells remain viable in the capsules. To assess possible effects of the capsule material in

a

b

c

d

Figure 2 Encapsulation of cytochrome P450 2B1-expressing cells in cellulose sulphate. (a) Encapsulated CYP2B1-transfected cells after double-staining and confocal laser microscopy. Green fluorescence indicates living cells and red fluorescence dead cells (×250 magnification). (b) Empty capsules injected into the pancreas of a nude (nu/nu) mouse (×200 magnification). (c) High power magnification (×400) of (b) to demonstrate lack of an inflammatory response. (d) Empty capsules injected into the pancreas of a Balb/c mouse pancreas (×400 magnification). All sections are shown 7 days after capsule implantation.


mice, empty (Figure 2b, c and d) and cell-filled (data not shown) capsules were injected orthotopically into the pancreas of both nude and immunocompetent Balb/c mice. Tissue reaction in both situations was minimal demonstrating few granulocytes and lymphocytes surrounding the capsule. No pancreatitis could be observed on the morphological level nor did the animals express signs of illness.

Treatment of pancreatic tumours To investigate the effects of local CYP2B1 expression on ifosfamide treatment of preformed tumours, 1 × 106 PaCa-44 human pancreatic tumour cells were injected subcutaneously into nude mice. PaCa-44 cells were chosen rather than PANC-1 because they give more aggressive and invasive tumours and are an established model for pancreatic carcinoma.16,17 When the resultant tumours had reached a size of 1 cm3, either 1 × 106 CYP2B1 expressing cells or 20–40 capsules carrying CYP2B1 cells were injected into the tumours and the animals treated systemically with ifosfamide. As expected, due to endogenous conversion of ifosfamide by the liver, tumour growth was inhibited in mice that received ifosfamide treatment, regardless of the presence or absence of CYP2B1 cells, compared with all of the nontreated groups. However, tumour reduction was most pronounced in mice that received encapsulated CYP2B1expressing cells by injection into the tumour and subsequent treatment with ifosfamide (Table 1 and Figure 3). Indeed, a complete disappearance of macroscopic and microscopic tumour growth was observed in four cases (Figures 3 and 4). Along with marked necrosis of the tumours (in comparison with the 10% background necrosis found in nontreated tumours), the BrdU labelling index was also dramatically reduced in the group treated with encapsulated CYP2B1 cells (Figure 5a and b), indicating that any remaining tumour cells were not actively dividing. In one particular animal, although a nodule was macroscopically visible, it displayed a complete cystic degeneration indicating eradication of the tumour. Even though the ifosfamide dosage was identical in all of the treated mice, animals treated with encapsulated cells also appeared healthier compared with those injected directly with CYP2B1 cells which at best showed

1073

a

b

c

Table 1 Experimental groups of nude mice and results after treatment with ifosfamide Group CYP2B1 cells

1 2 3 4 5 6 a

Treatment Number CR PRa % Mean of and necrosisb animals CR

No None Yes/naked None Yes/encapsulated None No Ifosfamide Yes/naked Ifosfamide Yes/encapsulated Ifosfamide

11 6 3 10 11 22

0 0 0 0 0 4

0 0 0 3 3 12

10 10 23 22 15 45

Partial responders (PR) were scored as mice with tumours of a diameter of 1.5 cm or less at completion of treatment regimen. Untreated animals scored tumour sizes of, on average, 2.5– 3.0 cm at completion. b Complete responders (CR) with either 100% necrosis or no tumour visible were scored 100%.

Figure 3 Subcutaneously xenotransplanted pancreatic tumours with and without treatment with encapsulated CYP2B1-transfected cells and ifosfamide. (a) Untreated tumour after 3 weeks, no injected cells or capsules, and no ifosfamide treatment. (b) The right hand part of a tumour was injected with encapsulated CYP2B1-transfected cells. After 3 weeks systemic treatment with ifosfamide, this region of the tumour shows 100% necrosis resulting in a fluid-filled cyst (arrow), as confirmed by pathology. (c) Tumour injected with encapsulated CYP2B1-transfected cells and treated systemically with ifosfamide showing complete disappearance of the tumour (CR) with residual scar tissue (arrow).


1074

a

b

c

Figure 4 Histology of human pancreatic carcinomas established in nude mouse. (a) Untreated tumour. (b) Tumour injected with capsules demonstrating viable encapsulated CYP2B1-expressing cells within the capsule. (c) Encapsulated CYP2B1-expressing cells after 3 weeks treatment with ifosfamide demonstrating extensive necrosis of tumour cells surrounding the capsule.

only a partial response (Table 1). Although the PaCa44 cell line is refractory to apoptotic signals in cell culture, an analysis of the percentage of cells in the tumour undergoing apoptosis revealed a dramatic increase from around 20% (Figure 5c) to about 60% (Figure 5d) after injection of encapsulated CYP2B1-expressing cells followed by ifosfamide treatment.

Discussion Therapeutic local concentrations of the toxic metabolites of ifosfamide, phosphoramide mustard and acrolein, are only achieved in conventional chemotherapy at the expense of high systemic concentrations, since the liver is the normal site of conversion. High systemic concentrations lead to severe effects on nontarget organs, especially the haematopoietic system. Direct delivery of these activated metabolites after in vitro production is compromised by the short half-life of these compounds (苲30 min).24 We reasoned that local cell therapy combined with systemic ifosfamide application will allow higher local concentrations of the active metabolite to be maintained without systemic toxicity. Treatment of xenotransplanted human pancreatic tumours by in situ chemotherapy, utilizing cell therapy with genetically modified encapsulated cells, resulted in a complete tumour disappearance in about 20% of the animals and a reduction in the tumour burden in the remaining 80%. The de novo expression of genes in cells and the physical targeting of this expression by injection of cells in an immunoprotected environment25 into the tumour represents a novel approach, combining established chemotherapy with gene therapy. The encapsulated cells maintained vitality and enzyme activities in cell culture for at least 4 weeks. Microscopic examination of the tumours after treatment, however, did not always demonstrate the presence of the capsules. Since the capsules are most likely located in the center of necrotic areas within the tumour, this is possibly a result of their loss during cutting and processing when they could be washed out with the almost liquid necrotic tumour tissue. This is supported by the observation that

such injected capsules are easily found after injection into normal organs such as the pancreas (Figure 2) or in the cleared mammary fat pad.36,37 As demonstrated by the injection of empty or cell-filled capsules in the unaffected pancreas of both nude and immunocompetent mice, no tissue reaction or pancreatitis could be observed 7 days after injection. In the case of the pancreas, this is an important issue since the organ is very sensitive to manipulation and ischemia.26 Injection of adenovirus for gene therapy into the pancreatic duct, for instance, caused pancreatitis.27 Thus, the local application of such capsules seems feasible. In a recent study, the CYP2B1 gene was transfected directly into breast cancer cells. Treatment with either cyclophosphamide or ifosfamide resulted in cell killing and after implantation in nude mice, tumour regression.28 This approach demonstrated the power of local delivery of the enzyme converting the cytotoxic drug locally into its active form. However, the genetic modification of tumour cells, whilst an elegant method to show proof of principle, cannot easily be applied in the clinic. Delivery of CYP2B1-transfected, encapsulated cells is a feasible clinical approach involving no direct gene therapeutic intervention in the patient and thus is safe and easier to handle. We have addressed the safety issue of whether cells escape from the capsules in the injected tumourbearing animals using PCR with primers specific for the expression construct. No CYP2B1-specific PCR product was detected in any organ including the brain and blood, nor was a PCR product found in tumour material from around the capsule implantation area of mice that were not treated with ifosfamide (data not shown). The experiments described above were carried out in a nude mouse model, thus the tumour killing effects observed stem entirely from the chemotherapy. In the human situation, the vigorous antitumour therapeutic effect of the active metabolites may stimulate antitumour immune responses. Unfortunately we were not able to investigate this potential effect because: (1) we are using a human pancreatic cancer cell line and thus must use an immunodeficient mouse model; and (2) no spontaneous pancreatic cancer model is available in mice. However,


1075

a

c

b

d

Figure 5 Analysis of tumour cell proliferation. (a, b) BrdU-labeling of pancreatic tumour cell nuclei from nude mice after (a) no treatment or (b) injection of encapsulated CYP2B1-expressing cells and treatment with ifosfamide in a partial responder (PR). (c, d) FACS sorting of propidium iodide labelled tumour cells, (c) without treatment or (d) after injection of encapsulated CYP2B1-expressing cells and treatment with ifosfamide. The number of anaploid cells (yellow) is sharply decreased after treament and there is a concomitant rise in the number of apoptotic cells (blue). The number of cells in G0/G1 is indicated in red.

it may well be that in a human clinical trial setting the antitumour effect of this therapy may be increased by immune involvement. Future improvements to this novel therapy could address any residual systemic toxicity effects which would be further reduced in patients by the locoregional application of ifosfamide (and even the capsules) through the arterial route, an option not feasible in the nude mouse model. The encapsulated cells themselves will most likely be killed by the administration of ifosfamide if they are dividing, since DNA alkylation is much more cytotoxic for cells that are undergoing cell division.13,14 The life span of these cells may be considerably enhanced

by their irradiation, before or after encapsulation. Thus, the data presented here, possibly in combination with such improvements, may prove useful in therapeutic approaches for pancreatic cancer and other solid tumours in man.

Materials and methods Construction of the cytochrome P450 2B1 expression construct The plasmid pcDNA3 (Invitrogen, NV Leek, The Netherlands) was digested with XhoI–XbaI and the


1076

were isolated 14 days later and tested for the presence and activity of the vector.

Figure 6 Box and Whiskers analysis of tumour necrosis observed in untreated mice (−/−), mice receiving nonencapsulated CYP2B1-expressing cells but no ifosfamide treatment (ne/−), mice treated only with ifosfamide (−/ifo), mice receiving nonencapsulated CYP2B1 expressing cells and ifosfamide (ne/ifo) or mice receiving encapsulated CYP2B1 expressing cells and ifosfamide (e/ifo). In a one-way analysis of variance (ANOVA), a significant variation was observed between the medians of the e/ifo group and the others (P ⬍ 0.05).

resulting fragments dephosphorylated using calf intestine phosphatase. The DNA of the vector backbone was purified by separation on a 1% agarose gel, excision and preparation using the Qiaquick protocol (Qiagen, Du¨sseldorf, Germany). After ethanol precipitation the DNA was resuspended in water. The cytochrome P450 2B1 carrying plasmid, pSW1,29 was digested with XhoI and XbaI to yield two fragments and the 1.5 kb fragment containing the rat cytochrome P450 2B1 cDNA,30 was excised and eluted using the Qiaquick DNA extraction protocol (Qiagen), ethanol precipitated and resuspended in water. 8.3 fmols of the pcDNA3 backbone and 24.8 fmols of the XhoI–XbaI fragment of pSW1 were mixed together and ligated for 1 day at 12°C using T4-ligase (Boehringer, Mannheim, Germany). The ligase was inactivated at 65°C for 10 min and the DNA butanol precipitated with a 10-fold volume of butanol. The precipitated DNA was resuspended in water and electroporated into DH10B-bacteria (Gibco, Eggenstein, Germany). Ampicillin-resistant colonies were selected, DNA-prepared and confirmed by restriction enzyme digestion as well as by sequencing.

Lipofection Before the day of transfection, 3 × 106 feline kidney cells18 were seeded into 100 mm dishes. On the day of transfection, 4 ␮g of the CYP2B1-expressing plasmid was mixed with 100 ␮l serum-free medium. In parallel, 15 ␮l Lipofectamine (Gibco) was mixed with 100 ␮l serum-free medium. The plasmid containing solution was added to the Lipofectamine mix and incubated for 45 min. After 35 min the cells were washed once with 2 ml serum-free medium. 800 ␮l of serum-free medium was added to the lipofection mix and the resulting 1 ml was put on to the prepared cells. After 6 h, 1 ml DMEM with Glutamax (Gibco) with 10% FCS were added. The next day the cells were trypsinised and diluted 10-fold and seeded on a 100-mm dish. After 24 h, the cells were given DMEM medium containing 400 ␮g/ml G418. Resistant clones

Cytochrome P450 enzymatic assay The expression of biologically active CYP2B1 in the transfectants was determined using a biochemical assay, which is specific for the cytochrome P450 isoform 2B1.19 Before the day of measurement, different amounts of cells (2.8 × 106, 5.5 × 105, 5.5 × 104) were seeded into a 3-cm dish. On the day of testing the cells were washed with phosphate-buffered saline (PBS; Gibco) and overlayed with 500 ␮l serum-free medium containing 15 ␮m 7-pentoxy-resorufin (Sigma, Deisenhofen, Germany) and 10 ␮m dicumarol (Sigma). Dicumarol inhibits the cellular diaphorase which would otherwise inactivate resorufin. After a 30-min incubation period at 37°C, 375 ␮l of the supernatant was mixed with 125 ␮l 0.1 mm sodium acetate pH 4.5 containing 75 Fishman units of ␤-glucuronidase per 600 Roy units of arylsulphatase (Boehringer). These two enzymes hydrolyse resorufin conjugates which would otherwise not be detected in the assay. The solution was incubated for 2 h at 37°C and the reaction stopped by adding 1 ml pure methanol. Precipitated proteins were pelleted at 3600 g and the amount of produced resorufin was measured with a fluorometer at 530 nm excitation and 590 nm emission. A standard curve was produced using different amounts of purified resorufin (Sigma). Cytochrome P450 2B1 functional in vitro assays To test the suicide effects of ifosfamide on CYP2B1 expressing cells, 4 × 104 cells were seeded into a 3-cm dish. After overnight incubation in DMEM containing Glutamax (Gibco) and 10% FCS, ifosfamide (200 mm) was added to a final concentration of 0.25–5 mm. After 5 additional days, the cells were trypsinized and an aliquot mixed with the same amount of 0.4% Trypan blue in PBS, incubated for 1 min and the living, nonstained cells counted. To test the bystander effect in the filter assay, the same number of CYP2B1-expressing and reporter (PANC1) cells were seeded on the first day, but separated by a membrane with 0.45 ␮m pores (Falcon, Heidelberg, Germany). The drug treatment was as described above. On the seventh day, the dead reporter cells were stained with trypan blue and the live cells counted. Encapsulation of cytochrome P450 expressing cells Capsules were produced as described.23 In short, 1 × 107 CYP2B1 transfected cells were suspended in 1 ml PBS (pH 7) containing 2–5% cellulose sulphate and 5% FCS (Gibco). The suspension was allowed to drop freely from an adjustable dispersion system by regulating the flow into a precipitation bath containing 3% polydiallyldimethyl ammonium in PBS. Capsule formation occurred within milliseconds followed by further constitution of an inner, more porous, layer for mechanical support. Capsules were washed twice with normal medium (RPMI) and then either taken into tissue culture or if empty stored in PBS at 4°C until further processing. To assess possible toxicity of the capsule material, empty capsules were injected orthotopically into the mouse pancreas, as described,31 both in nude and immunocompetent Balb/c mice (Charles River, Sulzfeld, Germany). Survival of cells was measured with the two-colour


Life&Dead Viability/Cytotoxicity Kit, which show a green fluorescence (calcein) for living cells and a red fluorescence (ethidium bromide; Sigma) for dead cells. Samples were processed according to the manufacturer’s recommendations. Encapsulated cells were analyzed with a confocal laser microscope (Zeiss, Jena, Germany).

Establishment and treatment of preformed pancreatic tumours in nude mice The human pancreatic carcinoma cell line PaCa-44 (ATCC), derived from a typical adenocarcinoma of moderate to poor differentiation (G2–3), was used. This cell line has been extensively characterized by us and others.17,32–34 The cell line carries mutations in codon 12 of the ras oncogene and mutations in exon 5/6 of the p53 tumour suppressor gene while RB1 is wild-type. PaCa44 cells were grown in RPMI with 10% FCS supplemented with penicillin and streptomycin (Gibco). Proliferating cells were used to establish tumours in the nude mouse. 1 × 106 cells were injected subcutaneously in the flanks of nude mice (CD-1 nu/nu; Charles River) in RPMI without supplements.17 Tumours were allowed to grow for 7 to 10 days. Experiments were started with tumours of comparable size (1 cm3). Several groups were defined (Table 1): (1) controls with no treatment and no injection, controls with injection of transfected fibroblasts (2) with and (3) without encapsulation but without ifosfamide treatment, and three treatment groups encompassing (4) no injection of cells, (5) injection of naked cells and (6) injection of encapsulated cells. 1 × 106 cells were suspended in 100 ␮l RPMI, filled into a 1-ml standard syringe (Braun, Melsungen, Germany) and injected through a 29 G needle (Microlance 3; Becton Dickinson, Fraga, Spain). Capsules were delivered through a 21 G needle (Microlance 3, inner diameter 0.6 mm) into the tumour. Approximately 20–40 capsules were delivered per injection. Animals were treated intraperitoneally every third day for 2 weeks with 100 mg/kg body weight ifosfamide (Holoxan; Asta Medica, Frankfurt-am-Main, Germany). At the same time, sodium 2mercaptoethanesulphonate (MESNA; Asta Medica) was administered i.v. at the same dosage (100 mg/kg body weight) via the tail vein. Tumour tissue was harvested from anesthetized animals after 3 weeks. The therapeutic effect was defined analogously to the terms complete response (CR), ie total disappearance of the tumour and partial response (PR), ie more than 50% reduction of tumour mass.35 The degree of tumour necrosis was determined semiquantitatively by two independent experimentors who had no previous knowledge of the treatment (ie blind). Necrosis was determined by examining sections under the microscope and counting necrotic areas using the grid in the eye piece. BrdU was administered at 100 mg/kg body weight i.p. 6 h before harvesting the tumour tissue from anaesthetized animals. The tumours were measured, cut in half and snap frozen immediately after removal from the animal. Sections of 4 ␮m were cut. Samples were processed according to the instruction of the kit (In situ Cell Proliferation Kit, AP; Boehringer). Cell nuclei were isolated from solid mouse tissues after mincing. 250 ␮l of trypsin buffer (solution A from Cycle TEST PLUS DNA Reagent Kit; Becton Dickinson, San Jose´, CA, USA) was added and gently mixed for 10 min

at room temperature. Trypsin inhibitor/RNase buffer (200 ␮l solution B, Cycle TEST) was added and the tube mixed again for 10 min at room temperature. Staining of nuclei was performed with 200 ␮l cold (4°C) propidium iodide (solution C, Cycle TEST) which was added to the nuclei suspension. After gently mixing for 10 min in the dark, the samples were incubated on ice for a further 15 min. The suspension was filtered through a 50 ␮m nylon mesh (Schleicher & Schuell, Dassel, Germany). Samples were stored on ice in the dark until measurement in the flow cytometer FACScan (Becton Dickinson, Heidelberg, Germany) using CELLQuest software (version 1.2.2., Becton Dickinson) for data aquisition. DNA histograms were analyzed with ModeFit LT software for Mac (version 1.01, Becton Dickinson). For every tumour sample, 20 000 nuclei were collected and stored in list mode files.

Acknowledgements We thank Jim Halpert (University of Arizona) for the cytochrome P450 cDNA, Dr Pohl (Asta Medica, Frankfurt) for helpful discussions about ifosfamide, Katrin Pu¨schel for performing the tissue culture, Hagen Pommerencke for suppying the Life&Dead pictures, Hartmut Stein for monitoring the nude mouse colony and Nora Sartori for her help in setting up the chemotherapy protocol. We would also like to thank Professor T Wagner (Medical University, Lu¨beck) for measuring the plasma concentration of ifosfamide after administration to mice and Gisela Sparmann for her input and support. This work was supported by Vaestfond grant from the Danish government to Bavarian Nordic Research Institute and EC grant BIO4-CT-0100.

References 1 Chabot GG. Factors involved in clinical pharmacology variability in oncology. Anticancer Res 1994; 14: 2269–2272. 2 Benard J, Bourhis J , Riou G. Clinical significance of multiple drug resisitance in human cancers. Anticancer Res 1990; 10: 1297–1302. 3 Kobayashi H et al. Acquired multicellular-mediated resistance to alkylating agents in cancer. Proc Natl Acad Sci USA 1993; 90: 3294–3298. 4 Rosewicz S, Wiedenmann B. Pancreatic carcinoma. Lancet 1997; 349: 485–489. 5 Abrams RA, Grochow LB. Adjuvant therapy with chemotherapy and radiation therapy in the managment of carcinoma of the pancreatic head. Surg Clin North Am 1995; 75: 925–938. 6 Howard TJ. Pancreatic adenocarcinoma. Curr Probl Cancer 1996; 20: 281–328. 7 Carmichael J et al. Phase II study of gemcitabine in patients with advanced pancreatic cancer. Br J Cancer 1996; 73: 101–105. 8 Ahlgren JD. Chemotherapy for pancreatic carcinoma. Cancer 1996; 78: 654–663. 9 Cerny T et al. Continuous 5-day infusion of ifosfamide with mesna in inoperable pancreatic cancer patients: a phase II study. J Cancer Res Clin Oncol 1991; 117: S135-S138. 10 Keizer HJ et al. Ifosfamide treatment as a 10-day continuous intravenous infusion. J Cancer Res Clin Oncol 1995; 121: 297–302. 11 Klapdor R et al. Determination of the tumour associated antigens CA 19–9 , CA 125 and CEA and the testing of drug sensitivity of gastrointestinal carcinomas transplanted to nude mice. In: Greten H, Klapdor R (eds). New Tumor-Associated Antigens. Georg Thieme: Stuttgart, 1986, pp 94–102.

1077


1078

12 Dirven H, van Ommen B, van Blaederen PJ. Glutathione conjugation of alkylating cytostatic drugs with a nitrogen mustard group and the role of glutathione S-transferases. Chem Res Toxicol 1996; 9: 351–360. 13 Wei MX et al. Experimental tumor therapy in mice using the cyclophosphamide-activating cytochrome P450 2B1 gene. Hum Gene Ther 1994; 5: 969–978. 14 Chen L, Waxman DJ. Intratumoral activation and enhanced chemotherapeutic effect of oxazaphosphorines following cytochrome P-450 gene transfer: development of a combined chemotherapy/cancer gene therapy strategy. Cancer Res 1995; 55: 581–589. 15 Elshami AA et al. Gap junctions play a role in the ‘bystander effect’ of the herpes simplex virus thymidine kinase/ganciclovir system in vitro. Gene Therapy 1996; 3: 85–92. 16 Klo¨ppel G, Lingenthal G, von Bulow M, Kern HF. Histological and fine structural features of pancreatic ductal adenocarcinomas in relation to growth and prognosis: studies in xenografted tumours and clinico-histopathological correlation in a series of 75 cases. Histopathology 1985; 9: 841–856. 17 Lo¨hr M et al. Human ductal adenocarcinomas of the pancreas express extracellular matrix proteins. Br J Cancer 1994; 69: 144–151. 18 Crandell RA, Fabricant CG, Nelson-Rees WA. Development, characterization, and virus susceptibility of a feline (Felis catus) renal cell line (CRFK). In Vitro 1973; 9: 176–185. 19 Donato MT, Gomez-Lechon MJ, Castell JV. A microassay for measuring cytochrome P450IA1 and P450IIB1 activities in intact human and rat hepatocytes cultured on 96-well plates. Anal Biochem 1993; 213: 29–33. 20 Wright JE et al. Analysis of 4-hydroxycyclophosphamide in human blood. Analyt Biochem 1995; 224: 154–158. 21 Freeman SM et al. The ‘bystander effect’: tumor regression when a fraction of the tumor mass is genetically modified. Cancer Res 1993; 53: 5274–5283. 22 Lieber M et al. Establishment of a continuous tumor-cell line (PANC1) from a human carcinoma of the exocrine pancreas. Int J Cancer 1975; 15: 741–747. 23 Stange J et al. Prolonged biochemical and morphological stability of encapsulated liver cells – a new method. Biomat Art Cells Immob Biotech 1993; 21: 343–352.

24 Sladek NE, Powers JF, Grage GM. Half-life of oxazaphosphorines in biological fluids. Drug Metab Dispos 1984; 12: 553–559. 25 Morris PJ. Immunoprotection of therapeutic cell transplants by encapsulation. Trends Biotech 1996; 14: 163–167. 26 Formela LJ, Galloway SW, Kingsnorth AN. Inflammatory mediators in acute pancreatitis. Br J Surg 1995; 82: 6–13. 27 Raper SE, DeMatteo RP. Adenovirus-mediated in vivo gene transfer and expression in normal rat pancreas. Pancreas 1996; 12: 401–410. 28 Chen L, Waxman DJ, Chen D, Kufe DW. Sensitization of human breast cancer cells to cyclophosphamide and ifosfamide by transfer of a liver cytochrome P450 gene. Cancer Res 1996; 56: 1331–1340. 29 Kedzie KM et al. Molecular basis for a functionally unique cytochrome P450IIB1 variant. J Biol Chem 1991; 266: 22515–22521. 30 Fuji-Kuriyama Y et al. Primary strucuture of a cytochrome P450: coding nucleotide sequence of phenobarbital-inducible cytochrome P-450 cDNA from rat liver. Proc Natl Acad Sci USA 1982; 79: 2793–2797. 31 Reyes G et al. Orthotopic xenografts of human pancreatic carcinomas acquire genetic aberrations during dissemination in nude mice. Cancer Res 1996; 56: 5713–5719. 32 Kalthoff H et al. Modulation of platelet-derived growth factor A- and B-chain/c-sis mRNA by tumor necrosis factor and other agents in adenocarcinoma cells. Oncogene 1991; 6: 1015–1021. 33 Lo¨hr M et al. Expression and function of receptors for extracellular matrix proteins in human ductal adenocarcinomas of the pancreas. Pancreas 1996; 12: 248–259. 34 Schmiegel W et al. Tumor necrosis factor ␣ and the epithelial growth factor receptor in human pancreatic cancer cells. Proc Natl Acad Sci USA 1993; 90: 863–867. 35 WHO. Handbook for Reporting Results of Cancer Treatment. 1979.

References added in proof 36 Saller RM, Gu¨nzburg WH, Salmons B. Microcapsules provide a novel alternative for systemic virus release; retrovirus vector slow release from packaging cell culture in a mouse gene therapy model. Gene Therapy 1995; 2 (Suppl. 1): S12. 37 Saller RM et al. Encapsulated cells producing retroviral vectors for in vivo gene therapy (submitted).